CLIN.CHEM.

26/1, 26-29 (1980)

Mechanism of Interference by Hemoglobin in the Determination of Total

Bilirubin. II. Method of Jendrassik-Grof
Bruce C. Shull,2 Helen Lees,1 and Philip K. Li3

Oxyhemoglobin in erythrocyte hemolysates interferes with kaline tartrate is used to convert neutral azobilirubin to the
the Jendrassik-Grof assay. Destruction of azobilirubin alkaline form. Increased absorbance of sample blanks of he-
occurs when oxyhemoglobin is oxidized to methemoglobin molyzed serum caused 2-4% negative error 1 h after alkalin-
during diazotization or to alkaline hematin with addition of ization (2).
alkaline tartrate. The most probable mechanism is by We compare the effects of hemolysis on the neutral and
oxidation with an agent such as hydrogen peroxide or a alkaline azobilirubin assays that involve caffeine reagent.
related species resulting from hemoglobin oxidation. Oxyhemoglobin and, to a lesser extent, methemoglobin are
Methemoglobin also appears to cause some destruction shown to be the interfering species. Kinetic studies demon-
of azobilirubin duringdiazotization. Methemoglobin forms strate the destruction of azobilirubin and the effects of he-
during diazotization because of reactions of oxyhemo- molysis on sample blanks.
globinwith both diazo reagent and nitrite ion.Formation
Materials and Methods
of methemoglobin is,therefore, more rapidinthe testthan
in the blank mixture and, under reaction conditions, its Instrumentation and reagents were as reported in the pre-
absorbance is less than that of oxyhemoglobin. This results ceding paper (Shull et al., this issue) with the following ad-
ditions:
in spectral interference when neutral azobilirubin is as-
sayed. Alkaline tartrate abolishes this spectral error by Instrumentation
causing rapid formation of alkaline hematin in both testand
blank. A Rotochem IIA/15 centrifugal analyzer (American In-
strument Co., Silver Spring, MD 20910) was used to perform
Hemolysis is known to cause negative error in bilirubin all bilirubin assays and rate studies.
assay by the Jendrassik-Grof method (1), although the error
is less than in the Malloy-Evelyn method (2,3). Michaelsson
Reagents
(2) found that azobilirubin destruction, after addition of al- Sulfanilic acid, 5.0 g/L in 0.18 mol/L HC1.
kaline tartrate, is the principal cause of the interference and Caffeine reagent. Dissolve 50 g of caffeine, 75 g of sodium
that adding ascorbic acid before alkaline tartrate stabilizes benzoate, and 125 g of sodium acetate trihydrate in distilled
azobilirubin. Although Micha#{235}lsson
claimed that ascorbic acid water at 50-60 #{176}C, and dilute to 1.0 L when cool.
could eliminate the effect of hemolysis, several papers still Working diazo reagent. Mix 10 mL of sulfanilic acid with
report hemolysis interference in various modifications of the 0.25 mL of sodium nitrite. Use within 1 h.
Jendrassik-Grof assay (4-6). Alkaline tartrate. Dissolve lOOgofNaOH and 285 gofdi-
Mon (6), using a nonalkalinized modification of the Jen- sodium tartrate dihydrate in distilled water at 50-60 #{176}C,and
drassik-Grof assay, found spectral error (high blanks) to be dilute to 1.0 L when cool.
the principal source of interference. This spectral error is Ascorbic acid. Dissolve 240 mg of ascorbic acid in 6.0 mL
thought to be due to more rapid formation of methemoglobin of distilled water. This is stable for 24 h if kept refriger-
in the test than in the blank. The absorbance of methemo- ated.
globin under reaction conditions is less than that of oxyhem- Potassium iodide, 235 mmol/L. This is stable for 24 h if
oglobin. Sims and Horn (7) found that nitrite ion was re- kept refrigerated.
sponsible for the greater rate of methemoglobin formation in Stabilized diazo reagent. Stabilized diazo reagent con-
the test. They recommend the addition of nitrite to the blanks taining no excess nitrite or sulfanilic acid was purchased from
as well as reading at the 525 nm isobestic point for oxyhemo- Union Carbide Corp., Rye, NY 10580. Each vial, containing
globin and methemoglobin. Mon (6) also used nitrite in the 14.5 mg of stabilized diazo salt, was reconstituted with 5.0 mL
blank to minimize spectral interference. The effectiveness of of 0.18 mol/L HC1.
nitrite blanks has been tested only at concentrations of he- Nitriteblank reagent. Mix 10 mL of 0.18 mol/L HC1 with
moglobin up to 1.4 g/L (6, 7). 0.25 mL sodium nitrite.
Spectral interference by hemoglobin is negligible when al-
Procedures
Jendrassik-Grof alkaline azobilirubin assay (9). The
transfer disc was loaded with sample (20 ML), water (80 zL),
1 Department of Medical Technology, State University of New York caffeine (200 ML), and diazo reagent (50 tL) into the large well
at Buffalo, Buffalo, NY 14214. and alkaline tartrate (150 iL) into the small well. After a
2 Present address: Dow Chemical Co., P.O. Box 68511 Indianapolis,

IN 46268. 10-mm incubation, 10 L of water or ascorbic acid was added

Departments of Pediatrics and Pathology, State University of to the large well, the run initiated, and absorbance measured
New York at Buffalo, Buffalo, NY 14214, and Clinical Laboratories, at 600 nm at various times, as indicated. Sulfanilic acid re-
The Childrens Hospital of Buffalo, Buffalo, NY 14222. Address placed working diazo reagent in sample blanks that were
correspondence to this author. otherwise identical. In some assays an initial 10-fold dilution
Presented by B.C.S. in partial fulfillment of requirements for the of sample was used (10 tL sample and 90 iiL of water). Unless
MS. degree, Department of Medical Technology, SUNY at Buffalo.
An account of experiments related to the method of Malloy-Evelyn
otherwise stated, however, assays were of fivefold initial
is described in the preceding article in this issue. sample dilution.
Received June 26, 1979; accepted Sept. 13, 1979. Neutral azobilirubinassay. The transfer disc was loaded

26 CLINICAL CHEMISTRY, Vol. 26, No. 1, 1980

with sample (40 ML), water (160 oL), and caffeine (400 tL) A.
1.00 -
into the large well and diazo reagent (100 zL) into the small
well. Absorbance was measured at 550 nm at various times,
as indicated. Assay temperature was 30 #{176}C.
Blanks were run
0.90
with hemolysate alone in human serum albumin, with either
sulfanilic acid or working diazo reagent, as indicated. In some
runs KI, 235 mmol/L, was used instead of distilled water as
0.80
the serum diluent.
Methemoglobin assay. Methemoglobin in the neutral U

azobilirubin reaction mixture was assayed by a modification B.

0 0.50
of the method of Evelyn and Malloy (10).Samples containing
hemolysate in human serum albumin were first incubated for
10 mm in the neutral azobilirubin assay mixture (blank and S
0.40
test). The absorbance at 630 nm was read before and after the
addition of excess KCN. The percent of methemoglobin was
calculated by comparison
which K3Fe(CN)6
with a matched
was used to complete
set of samples
the conversion
in
to I
0 2
#{149}
4 6 $ 10 12
methemoglobin.
IIEMOCLOSIN, g/L
Spectral absorbance scans. Samples of hemolysate in
human serum albumin were scanned vs reagent blanks from Fig. 1. The effect of added hemoglobin derivatives on Jen-
400-640 nm at various times after incubation in the neutral drassik-Grof assay
or alkaline azobilirubin assay mixture. Hemoglobin derivatives Added hemolysate. (#{149}):
added oxyhemoglobln, (0); added methemoglobin, ().
were identified by comparison with published spectra (11). Bilirubin concentration was 187 mg/L. Ascorbic acid was added and absorbances
read,at 600 nm, 30 s after alkalinization. A, fivefold initial sample dilution; B,
10-fold initial sample dilution
Results and Discussion
Hemoglobin Species in the Reaction Mixture Effect of Hemolysate on Standard Curves
Addition of hemolysate, 10 g/L, significantly (p <0.05)
Spectral absorbance scans of samples of hemolysate in
depressed the slope of the standard curve in the alkaline assay;
human serum albumin demonstrated progressive formation
but the intercept was unaffected (p > 0.05) (Figure 2). These
of methemoglobin in the neutral azobilirubin assay mixture.
data suggest chemical rather than spectral interference by
After 10 mm methemoglobin averaged 28% in the test sample
hemolysis. Pearlman and Lee (4) also found that hemolysis
vs 22% in the blank mixture containing sulfanilic acid. Use of
interference in the Jendrassik-Grof assay is proportional to
nitrite blank reagent increased the rate of methemoglobin
both bilirubin and hemoglobin concentration.
formation over that with sulfanilic acid, as previously reported
by Sims and Horn (7). However, stabilized diazo reagent Hemolysis Interference during Diazotization
further increased the rate of methemoglobin formation; thus
Absorbance significantly decreased (p <0.05) in proportion
the use of nitrite-containing blanks did not equalize rates of
to the amount of added hemoglobin, within 12 s of initiating
methemoglobin formation in test and blank at high concen-
the diazo reaction (Table 1). After 12.5 mm the error tripled.
trations of hemoglobin.
KI did not affect absorbances at 12 s but significantly (p <
Alkaline tartrate caused prompt formation of alkaline he-
0.05) lowered the error after 12.5 mm.
matin in both test and blank, thereby eliminating incorrect
A plot of absorbance vs time is given in Figure 3. Fading of
blank correction. This probably accounts for the observation
color after I mm was eliminated by using a working diazo
of Peaniman and Lee (4) that hemolysis causes greater inter-
ference in neutral azobilirubin assay than in the alkaline
assay.

Interfering Species
The decrease in test absorbance in the alkaline assay was
linearly related to added hemoglobin at five-fold initial sample
dilution (Figure 1). Regression lines relating test absorbance
to hemoglobin concentration did not differ significantly (p
> 0.05), whether hemolysate or oxyhemoglobin preparation
was added; methemoglobin, however, caused significantly less
(p <0.05) interference than hemolysate. The interference by
hemolysate and oxyhemoglobin was not significant, p > 0.05, 2

at 10-fold initial sample dilution, in agreement with Mi-

cha#{235}lsson
(2).Neither reduced glutathione, up to 100 mg/L,
or dialysate from hemolyzed erythrocytes caused interfer-
ence.
These experiments identified oxyhemoglobin as the prin- 0.10

cipal interfering species. Methemoglobin, which forms during

diazotization, also interferes but does not fully account for
hemolysis error. 1 20 40 60 80 100 20 40

These data also demonstrate that ascorbic acid does not RILIRUHIPI, .q/I.

abolish hemolysis error at initial sample dilutions less than

Fig. 2. The effect of added hemolysate on standard curves in the
10-fold. Variation of initial sample dilution probably explains Jendrassik-Grof assay
some conflicting reports about the extent of hemolysis error No hemolysate,(#{149});
10g/L hemolysate added, (0). Ascorbic acid was added
in the Jendrassik-Grof assay. and absorbances read, at 600 nm, 30 a after alkalinization

CLINICAL CHEMISTRY, Vol. 26, No. 1, 1980 27

 TESS ABSORBANE
Table 1. The Effect of Added Hemolysate on the
Jendrassik-Grof Neutral Azobilirubin Assay a 0.00
Reading SD of
time n Slop., b Slope. Sb
12 s 6 -2.47 X i0 0.42 X iO 0.85
12 s, KI addedb 6 -2.55 X io- 0.30 X iO
12.5 mm 6 -8.70 X io- 0.59 X iO B
C
12.5 mm, KI addedb 6 -6.60 X i0 0.40 X 1O o 0.20
HEMOGLOBIN BLANK

Addedhemoglobinwas 0-10 g/L and bllirubin 187 mg/L. Slope of line re-
lating absorbance to hemoglobin concentration was calculated by linear re- ABSOBBANE
0.15
gression. 0
b KI, 50 mmol/L final concentratIon, added as an antioxidant.
.*c.u-eao..

0.10
blank and including K! in the reaction mixture. KI did not
affect test absorbance when there was no added hemolysate.
Progressive spectral interference is evident from the plots of
0 2 4 6 $ 10 12
hemoglobin absorbance in the test and blank. Even with TIHE, BIN
correction of spectral interference, a constant rate of fade
remained, which was stabilized by KI. Fig.4.Effectofadded hemolysateon absorbancevs. time after
Destruction of bilirubin by hemoglobin before diazotization alkalinization in the Jendrassik-Grof assay
Test absorbancesread againstsample blank containingsulfanilic acid:
could account for the decrease in absorbance within 12 s of no hemoglobin; 0-0, 10 g/L hemoglobin; #{149}-#{149},
10g/L hemoglobin with
reaction time. However, preincubation in caffeine did not ascorbicacid in reaction mixtta-e.Hemoglobinabsorbancein thereactionmixture
increase the error of a sample with 10 g of added hemolysate read against water: 0-0, sulfanilic acid blank; #{149}-#{149},
working diazo blank
per liter. Direct spectrophotometry at 455 nm confirmed the
stability of bilirubin in caffeine reagent. This is in agreement irubin occurs in hemolyzed samples after alkalinization but
with Micha#{235}lsson (2),who found that caffeine has a stabilizing that ferric heme cannot be the oxidizing agent.
effect on bilirubin. As previously noted, alkaline tartrate eliminates spectral
Methemoglobin also caused progressive fading of azobil- interference. Hemoglobin absorbances in test and blank were
irubin after 1 mm but did not affect absorbance during the
stable, after an initial decline, for up to 12.5 mm (Figure 4).
first 12 s of the reaction.
Proposed Mechanism of Azobilirubin Oxidation
Hemolysis Interference afterAlkalinization
Destruction of azobilirubin in hemolyzed serum has been
Samples with added hemolysate showed fading after ad- demonstrated both during diazotization and after alkalin-
dition of alkaline tartrate, but this was prevented by adding ization. The stabilizing effect of reducing agents suggests that
ascorbic acid (Figure 4). Neither methemoglobin, to 10 g/L, the interference is oxidative. Under both conditions, the most
nor methemealbumin, toll g/L as hemoglobin, caused fading probable oxidizing agent would be peroxide or a related
of azobilirubin. These data suggest that oxidation of azobil- species formed when heme undergoes oxidation with con-
comitant reduction of 02. Ferric heme is known to function
TOOT ADSORRAIICE
as a pseudoperoxidase (12)and could, therefore, play a cata-
lytic role in azobilirubin oxidation.
0.80
Wallace and Caughey (13) have demonstrated two mech-
anisms for the oxidation of oxyhemoglobin, both of which
generate oxidizing species. In the presence of external one-
0.75 electron donors, such as nitrite ion, 02 is reduced by two
electrons to form peroxide and ferric heme. Under denaturing
0.70
conditions, heme itself may act as the external electron donor.
Nucleophiles, such as azide ion, cause reductive displacement
41000L.OB0U BLANK
*890 88*9 CE to yield superoxide anion and ferric heme. Mirsa and Frido-
0.65 vich (14) found both superoxide and peroxide formation
during autoxidation of oxyhemoglobin, and demonstrated
co-oxidation of epinephrine by these species.
0.60
Methemoglobin reacts with a number of reducing agents
to form hemoglobin (15). This may account for the fading of
0.55
azobilirubin during diazotization with added methemoglobin.
However, as noted previously, oxyhemoglobin is the principal
interfering species. Ferric heme, added as methemoglobin or
1#{176} methemealbumin, does not cause the destruction of azobil-
0 2 4 6 $ 10 12
irubin after alkalinization.
TONI. 900

Fig. 3. Effect of hemolysate on absorbance vs time in the Jen- References

drassik-Grof neutral azobilirubin assay 1. Watson, D., A note on the haemoglobin error in some non-pre-
Test absorbance: D-f>, no added hemoglobin; 0-0, 10 g/L hemoglobin vs cipitation diazo-methods for bilirubin determinations. Clin.Chim.
suifanilic acid blank; #{149}-#{149},
10 g/L hemoglobin vs working diazo blank; Acta 5,613 (1960).
10 g/L hemoglobin and 50 mmoi/L Ki vs working diazo blank. Bilirubin con- 2. Micha#{235}lsson,
M., Bilirubin determination in serum and urine.
centration was 187 mg/L. Sulfanhlic acid blank contained 10 g of hemoglobin
per liter; working diazo blank was test sample wIth 0 bilirubin and 10 g of he-
Scand.J. Clin.Lab.Invest.13, Suppl. 56 (1961).
moglobin per liter. Absorbance due to hemoglobin In these reaction mixtures 3. Micha#{235}lsson,
M., Nosslin, B., and Sj#{246}lin,
S., Plasma bilirubin de-
read against water: 0-0. sulfanilic acid blank; #{149}-#{149},
working diazo blank termination in the newborn infant. Pediatrics 35, 925 (1965).

28 CLINICALCHEMISTRY,Vol. 26, No. 1, 1980

4. Pearlman, F. C., and Lee, R. T. Y., Detection and measurement hemoglobin, methemoglobin, and sulfhemoglobin in a single sample
of total bilirubin in serum, with use of surfactants as solubilizing of blood. J. Biol. Chem. 126,655 (1938).
agents. Clin. Chem. 20,447 (1974). 11. Antonini, B., and Brunori, M., Hemoglobin and Myoglobin in
5. Waud, W. R., Cross, R. E., and Savory, J., Selection of a total bi- Their Reactions with Ligands. North Holland Publishing Co., Am-
lirubin method for pediatric specimens: Problem of hemolysis. Clin. sterdam, 1971.
Chem. 23, 1165 (1977). 12. Clinical Chemistry, Principles and Technics, 2nd ed., R. J.
6. Mori, L, Modified Jendrassik-Grof method for bilirubins adapted Henry, D. C. Cannon, and J. W. Winkelman, Eds., Harper and Row,
to the Abbott Bichromatic Analyzer. Gun. Chem. 24, 1841 (1978). Hagerstown, MD, 1974, p 1140.
7. Sims, F. H., and Horn, C., Some observations on Powells method 13. Wallace, W. J., and Caughey, W. S., Mechanism for the autox-
for the determination of serum bilirubin. Am. J. Clin. Pat ho!. 29,412 idation of hemoglobin by phenols, nitrite, and oxidant drugs. Per-
(1958). oxide formation by one electron donation to bound dioxygen. Bio-
chem. Biophys. Res. Commun. 62,561(1975).
8. Shull, B.C., Lees, H., and Li, P. K., Mechanism of interference by
14. Mirsa, H. P., and Fridovich, I., The generation of superoxide
hemoglobin in the determination of total bilirubin. I. Method of
radical during the autoxidation of hemoglobin. J. Biol. Chem. 247,
Malloy-Evelyn. Clin. Chem., 26,22-25(1980).
6960 (1972).
9. Nosslin, B., The direct diazo reaction of bile pigments in serum. 15, Castro, C. E., Wade, R. S., and Belser, N. 0., Conversion of oxy-
Scand.J.Clin.Lab. Invest. 12, Suppl. 49 (1960). hemoglobin to methemoglobin by organic and inorganic reductants.
10. Evelyn, K. A., and Malloy, H. T., Microdetermination of oxy- Biochemistry 17, 225 (1978).

CLINICALCHEMISTRY, Vol. 26, No. 1, 1980 29